2850
Chem. Mater. 2001, 13, 2850-2863
Photoluminescence and Magnetism in the New Magnetic Semiconductors: K2Cd3(1-x)Mn3xS4 Enos A. Axtell, III,† Jason Hanko,† Jerry A. Cowen,‡,§ and Mercouri G. Kanatzidis*,† Departments of Chemistry and Physics and Astronomy, and Center for Fundamental Materials Research, Michigan State University, East Lansing, Michigan 48824-1322 Received January 18, 2001. Revised Manuscript Received May 22, 2001
A new family of magnetic semiconductors, K2Cd3(1-x)Mn3xS4, prepared from molten K2Sx is reported. For x e 0.6, the compounds crystallize in the K2Cd3S4 structure type. The cell volumes of the series obey Vegard’s Law for solid solutions. The compounds range in color from yellow through orange brown to red and possess platelike morphologies. A structural transformation occurs between x ) 0.7 and x ) 0.9. Red plates of K2Cd0.4Mn2.6S4 (x ) 0.87) crystallize in the space group Ccca with a ) 5.9217(8) Å, b ) 13.531(1) Å, c ) 11.0696(8) Å, and V ) 887.0(1) Å3. The end member, K2Mn3S4, crystallizes in the space group Ccca with a ) 11.0637(7) Å, b ) 26.830(2) Å c ) 5.8180(4) Å, and V ) 1727 Å3. For x e 0.6, the compounds form as (M3S4)n2n- layers, interspersed by K+ cations. These layers are composed only of M3S42- units shaped like truncated cubes. When x ) 0.87, a new structure related to that of Cs2Mn3S4 is observed. The end member of the series (i.e., x ) 1) K2Mn3S4 is revealed to have a related yet surprising structure. The compounds display room-temperature band gaps ranging from 2.79 to 2.92 eV. Absorptions in the mid-gap region, due to Mn2+-based d-d transitions, are also observed. The compounds are strongly emissive in the visible spectrum at room temperature, with the light emission shifted to the red immediately upon the introduction of small amounts of manganese. The magnetic properties of these compounds evolve from classical paramagnetic to spin glass as a function of x. For x g 0.05, the compounds deviate from Curie-Weiss behavior. Antiferromagnetic coupling is observed in all cases. With very high Mn content transitions to magnetically complex phases are observed with possible spin-glass-like behavior.
Introduction Compounds that display both semiconducting and magnetic properties are interesting experimentally, theoretically, and technologically.1 Over the last several decades, interest has focused on the so-called II-VI compounds of the type M1-xMnxQ (M ) Zn, Cd, Hg; Q ) S, Se, Te).2 These dense, three-dimensional compounds display tetrahedral coordination about both the metal and the nonmetal and their crystallographic, optical, and magnetic properties have been studied thoroughly. Another facet of magnetic semiconductor research is the study of manganese-doped, II-VI quan* To whom correspondence should be addressed. † Department of Chemistry. ‡ Department of Physics and Astronomy. § Deceased. (1) Ohno, H.; Chiba, D.; Matsukura, F.; Omiya, T.; Abe, E.; Dietl, T.; Ohno, Y.; Ohtani, K. Nature 2000, 408, 944-946. (b) Ohno, H. Science 1998, 281, 951-956. (c) Twardowski, A. Mater. Sci. Eng., B 1999, 63, 96-102. (d) Bourgognon, C.; Tatarenko, S.; Cibert, J.; Gilles, B.; Marty, A.; Samson, Y. Appl. Phys. Lett. 1999, 75, 2818-2820. (e) Dietl, T.; Ohno, H.; Matsukura, F.; Cibert, J.; Ferrand, D. Science 2000, 287, 1019-1022. (f) Lidiard, A. B. Rep. Prog. Phys. 1954, 27, 201. (2) Diluted Magnetic (Semimagnetic) Semiconductors; Aggarwal, R. L., Furdyna, J. K., von Molnar, S., Eds.; Materials Research Society Symposia Proceedings; Materials Research Society: Pittsburgh, PA, 1987; Vol. 89, and references therein. (b) Bindilatti, V.; ter Haar, E.; Oliveira, N. F.; Shapira, Y.; Liu, M. T. Phys. Rev. Lett. 1998, 80, 54255428. (c) Levy, L.; Ingert, D.; Feltin, N.; Pileni, M. P. Adv. Mater. 1998, 10, 53. (d) Lunn, B.; Davies, J. J. Semicond. Sci. Technol. 1990, 5, 1155-1160.
tum-sized particles as for example the Cd1-xMnxSe system.3,4 The particles display increasing band gaps with increasing manganese content, and spin glass behavior is observed for low concentrations of Mn.5 Other dilute magnetic semiconductor systems with interesting magnetic properties include Zn2-xMnxP2S66 and Cd2-xFexP2S67 and Zn1-xMnxGa2Se48 with 0 < x < (3) Kim, K. W. Ph.D. Dissertation, Michigan State University, East Lansing, MI, 1993. (b) Kim, K.-W.; Cowen, J. A.; Dhingra, S.; Kanatzidis, M. G. In Chemical Processes in Inorganic Materials: Metal and Semiconductor Clusters and Colloids; Bradley, J. S., Persans, P. D., Schmid, G., Chianelli, R. R., Eds.; Materials Research Society Symposia Proceedings; Materials Research Society: Pittsburgh, PA, 1992; Vol. 272, pp 27-33. (b) Levy, L.; Feltin, N.; Ingert, D.; Pileni, M. P. Langmuir 1999, 15, 3386-3389. (c) Feltin, N.; Levy, L.; Ingert, D.; Pileni, M. P. J. Phys. Chem. B 1999, 103, 4-1. (d) Pileni, M. P. Cryst. Res. Technol. 1998, 33, 1155-1186. (4) Levy, L.; Ingert, D.; Feltin, N.; Pileni, M.-P. Adv. Mater. 1998, 10, 53. (5) Yang, Y. Q.; Keesom, P. H.; Furdyna, J. K.; Giriat, W. J. Solid State Chem. 1983, 49, 20-24. (b) Twardowski, A.; Denissen, C. J. M.; Dejonge, W. J. M.; Dewaele, A.; Demianiuk, M.; Triboulet, R. Solid State Commun. 1986, 59, 199-203. (c) Bandaranayake, R. J.; Lin, J. Y.; Jiang, H. X.; Sorensen, C. M. J. Magn. Magn. Mater. 1997, 169 (3), 289-302. (d) Bandaranayake, R. J.; Smith, M.; Lin, J. Y.; Nang, H. X.; Sorensen, C. M. IEEE Trans. Magn. 1994, 30, 4930-4932. (6) Chandrasekharan, N.; Vasudevan, S. Phys. Rev. B-Condens. Matter Mater. Phys. 1996, 54, 14903-14906. (b) Goossens, D. J.; Hicks, T. J. J. Phys.: Condens. Matter 1998, 10, 7643-7652. (7) Leaustic, A.; Riviere, E.; Clement, R.; Manova, E.; Mitov, I. J. Phys. Chem. B 1999, 103, 4833-4838. (8) Moron, M. C.; Campo, J.; Palacio, F.; Attolini, G.; Pelosi, C. J. Magn. Magn. Mater. 1999, 197, 437-439.
10.1021/cm0100422 CCC: $20.00 © 2001 American Chemical Society Published on Web 08/24/2001
Photoluminescence and Magnetism in K2Cd3(1-x)Mn3xS4
1. Antiferromagnetic- as well as spin-glass-like behavior was observed depending on the Mn content and the cation distribution. Dilute magnetic semiconductor systems are now attracting considerable interest for their potential in the field of spintronics where electrically controlled ferromagnetism in manganese compound semiconductors opens prospects for tailoring magneticand spin-related phenomena in semiconductors and enabling spin-based devices.1 Recently, we described the new family of compounds A2Cd3Q4 (A ) K, Rb; Q ) S, Se, Te)9 that feature unique [Cd3S4]2- layers with cuboidal Cd3S4 cluster. The unique, low-dimensional structure and the possible superexchange paths in the K2Cd3S4-type compounds have led us to a structural, optical, and magnetic study of K2Cd3(1-x)Mn3xS4 solid solutions, with 0 < x < 1. Theoretically, Mn atoms would replace Cd atoms in the (M3S4)n2n- layers. Each truncated cuboidal fragment in this layered structure contains three metal atoms connected together with S2- ions. In each truncated cube the metals form an almost equilateral triangle with M-M distances of ∼3.2 Å. From a magnetic point of view the triangular geometry presents a degenerate ground state which could give rise to unusual magnetic phenomena.10 Thus, increasing amounts of manganese could result in the phenomenon of “frustrated spins,” leading to spin-glass-like behavior.11 Other possible long-range ordering could include ferromagnetism, antiferromagnetism, or ferrimagnetism, as is observed in Cd2-xMnxP2S6. In contrast to the behavior found in the systems described here, a similar study of solid solutions in the system Cs2Zn3(1-x)Mn3xS4, which does not possess truncated cuboidal clusters, indicated only antiferromagnetic ordering.12 Here, we report in detail the synthesis of K2Cd3(1-x)Mn3xS4 solid solutions and describe the evolution of the structural, optical, photoluminescent, and magnetic properties of these compounds as a function of Mn. Interestingly, structural transitions are observed as a function of x. The compounds display both strong red emission, even at room temperature, and interesting magnetic behavior. We also describe the structure of K2Mn3S4, which defines the end member of the series (i.e., x ) 1) and has a surprising structure. Experimental Section K2Cd3(1-x)Mn3xS4 Solid Solutions. All the K2Cd3(1-x)Mn3xS4 solid solutions were produced by reacting stoichiometric amounts of Cd and Mn metals in excess K2S3. The details of the synthesis of each material are summarized in Table 1. A charge of 0.331 g (3.0 mmol) of K2S, appropriate amounts of Cd metal and Mn metal, (9) Axtell, E. A.; Liao, J.-H.; Kanatzidis, M. G. Chem. Eur. J. 1996, 2, 656-666. (10) Suzuki, I. S.; Suzuki, M. J. Phys.: Condens. Matter 1999, 11, 521-541. (b) Cava, R. J.; Ramirez, A. P.; Huang, Q.; Krajewski, J. J. J. Solid State Chem. 1998, 140, 337-344. (c) Schiffer, P.; Daruka, I. Phys. Rev. B 1997, 56, 13712-13715. (d) Mentre, O.; Dhaussy, A. C.; Abraham, F.; Steinfink, H. J. Solid State Chem. 1997, 130, 223-233. (e) Zhang, W. M.; Saslow, W. M.; Gabay, M.; Benakli, M. Phys. Rev. B 1993, 48, 10204-10216. (f) Reimers, J. N.; Dahn, J. R.; Greedan, J. E.; Stager, C. V.; Liu, G.; Davidson, I., Vonsacken, U. J. Solid State Chem. 1993, 102, 542-552. (g) Reimers, J. N.; Greedan, J. E.; Bjorgvinsson, M. Phys. Rev. B 1992, 45, 7295-7306. (11) Binder, K.; Young, A. P. Rev. Mod. Phys. 1986, 58, 801-976. (12) Heming, M.; Lehmann, G. Z. Naturforsch A 1983, 38, 149153. (b) Bronger, W.; Balkhardtdegen, H.; Muller, P. Z. Kristallogr. 1988, 182, 47-49.
Chem. Mater., Vol. 13, No. 9, 2001 2851 Table 1. Conditions for the Preparation of K2Cd3(1-x)Mn3xS4 Solid Solutions x (nominal)
K2S/S (mmol)
Cd (mmol)
Mn (mmol)
obsd composition by EDS analysis
0.0 0.05 0.1 0.2 0.4 0.6 0.87 1.0
3/6 3/6 3/6 3/6 3/6 3/6 3/6 3/6
6g 5.7 5.4 4.8 3.6 2.4 0.6 0
0 0.3 0.6 1.2 2.4 3.6 5.4 6
K2Cd3S4 K2Cd2.76Mn0.19S4 K2Cd3.3Mn0.4S4 K2Cd2.5Mn0.6S4 K2Cd1.8Mn1.2S4 K2Cd1.6Mn1.4S4 K2Cd0.2Mn3.0S4 K2Mn3S4
Table 2. Lattice Parameters and Space Groups of the K2Cd3(1-x)Mn3xS4 Solid Solutions x
a (Å)
b (Å)
0.0 0.2 0.4 0.6 0.87 0.88 1.0
13.880(4) 13.884(6) 13.863(4) 13.851(4) 5.9217(8) 5.890(1) 11.0637(7)
c (Å)
10.247(3) 6.608(1) 10.224(4) 6.542(3) 10.184(1) 6.480(1) 10.139(2) 6.428(2) 13.531(1) 11.070(1) 13.489(1) 11.066(1) 26.830(2) 5.818(1)
volume (Å3) space group 939.8(7) 928.6(7) 914.8(3) 902.7(4) 887.0(1) 879.4(2) 1727(1)
Pnma Pnma Pnma Pnma Ccca Ccca Ccca
and 0.192 g (6.0 mmol) of elemental S was loaded into an alumina thimble. The thimble was then sealed inside a 13-mm-o.d. × 11-mm-i.d. carbon-coated silica tube. The tube was placed into a computer-controlled furnace, heated at 800 °C for 2 days, and then cooled at 2 °C/h to 700 °C and at 25 °C/h to 50 °C. The excess of K2S3 promoted crystal growth, prevented the formation of Cd1-xMnxS, and was washed away with methanol. The K2Cd3(1-x)Mn3xS4 solid solutions were isolated with methanol under a nitrogen atmosphere. A 75% yield, based on Cd and Mn, is typical. The compounds are insoluble in H2O and common organic solvents, but are sensitive to long exposure (several days) to air. The elemental compositions of all materials were verified by semiquantitative microprobe analysis on several crystals from each batch; see Table 1. K2Mn3S4. A charge of 0.331 g (3.0 mmol) of K2S, 0.330 g (6.0 mmol) of Mn metal, and 0.192 g (6.0 mmol) of elemental S was loaded into an alumina thimble, which was sealed inside a carbon-coated silica tube. The tube was heated at 800 °C for 2 days and then cooled at 2 °C/h to 700 °C and at 25 °C/h to 50 °C. The product was isolated as above. The compound is insoluble in H2O and common organic solvents, but is sensitive to air exposure for several days. A yield of 60%, based on Mn, is typical. Crystallographic Studies Single-Crystal X-ray Diffraction. K2Cd3(1-x)Mn3xS4 (x e 0.6). Because for x e 0.6 the compounds have the structure of K2Cd3S4 (x ) 0), only the unit cells were determined at 23 °C; see Table 2. For the sample with x ) 0.6 data collection and crystallographic refinement were performed to verify the structure; however, these details are not reported here. K2Cd3(1-x)Mn3xS4 (x ) 0.87, 0.88). Single-crystal X-ray diffraction data for K2Cd0.39Mn2.61S4 (x ) 0.87) were collected on a Rigaku AFC6S four-circle diffractometer (Mo KR radiation, graphite monochromated) at 23 °C. Cell parameters were a ) 5.922(1) Å, b ) 13.531(1) Å, c ) 11.070(1) Å, Ccca, and final R/Rw ) 0.046/0.089. The crystal was stable, as judged by three check reflections measured every 150 reflections throughout the data set. The structure was solved by the direct methods routine of SHELXS-8613 and refined by the full-matrix least(13) Sheldrick, G. M. In Crystallographic Computing 3; Sheldrick, G. M., Kruger, C., Doddard, R., Eds.; Oxford University Press: Oxford, England, 1985; pp 175-189.
2852
Chem. Mater., Vol. 13, No. 9, 2001
Axtell et al.
Table 3. Summary of Crystallographic Data for K2Cd3(1-x)Mn3xS4 formula
K2Cd0.35Mn2.65S4
FW (g/mol) color/habit dimensions (mm) a (Å) b (Å) c (Å) V (Å3) space group, Z dcalc (g/cm3) µ(Mo KR) (cm-1) radiation temperature (°C) scan type 2θmax (deg) min./max. absorption coeff. for ψ-scan correction reflections measured no. of reflections with I > 3σ(I) no. variables R/Rwa
K2Mn3S4
391.51 red/irreg 0.30 × 0.22 × 0.12 5.890(1) 13.489(1) 10.066(1) 879.4(2) Ccca, 4 2.957 63.9 Mo KR (λ ) 0.71069 Å) -100 2θ/ω 55 0.6/1.1
371.26 red/irregular plate 0.46 × 0.26 × 0.08 11.0637(7) 26.830(2) 5.8180(4) 1727.0(2) Ccca, 8 2.856 61.55 Mo KR (λ ) 0.71069) Å -100 2θ/ω 55 0.28/0.76
2532 547
3612 787
23 0.046/0.14
47 0.081/0.20
a R ) Σ(|F | - |F |)/Σ|F |R ) {Σw(|F 2| - |F 2|)2/Σw|F 2|2}1/2w ) o c o w o c o 1/σ2(F).
Table 4. Fractional Atomic Coordinates (×104), Ueqa Values and Site Occupancies for K2Cd0.35Mn2.65S4 with Estimated Standard Deviations in Parentheses Cd1 Mn1 Mn2 K S1
Table 5. Fractional Atomic Coordinates (×104), Ueqa Values and Site Occupancies for K2Mn3S4
x
y
z
Ueqa
occupancy
0 0 0 -7500 2211(2)
2500 2500 2500 0 1430(1)
5398(7) 5209(3) 2500 1289(1) 3835(1)
31(3) 9(1) 5(1) 19(1) 12(1)
0.18 0.82 1.0 1.0 1.0
a U eq is defined as one-third of the trace of the orthogonalized Uij tensor.
squares techniques of the TEXSAN package of crystallographic programs.14 An empirical absorption correction based on ψ-scans was applied to the data set, followed by a DIFABS correction to the isotropically refined structure as recommended.15 The structure was modeled satisfactorily with 13% Cd and 87% Mn on the two metal sites. All atoms were eventually refined anisotropically. To check for reproducibility, another sample (from a different synthesis batch) was also refined and gave x ) 0.88. The data were collected with a SMART Platform CCD diffractometer as described below. The structural model was identical to that of the x ) 0.87 sample. Because of the substantially identical results obtained, the complete parameters and details of the structure solution and refinement are given only for K2Cd0.35Mn2.65S4; see Table 3. The coordinates of all atoms, isotropic equivalent temperature factors, and their standard deviations are given in Table 4. K2Mn3S4. The structure of this compound was recently described by Bronger et al. albeit we believe in an incorrect unit cell and space group.16 K2Mn3S4 was reported to crystallize in the monoclinic space group P2/c, with a ) 7.244(2) Å, b ) 5.822(1) Å, c ) 11.018(5) Å, β ) 112.33(3)°, and Z ) 2. The [Mn3S4] layers were described to be of the [PbO] type with disordered vacancies in the Mn sites. Our results differ from their conclusions in that we have observed a superstructure over that described by the monoclinic structure. The crystals of K2Mn3S4 did not diffract well and at first we were not able to collect good diffraction data for this (14) TEXSAN: Single-Crystal Structure Analysis Software, Version 5.0; Molecular Structure Corp.: The Woodlands, TX, 1981. (15) Walker, N.; Stuart, D. Acta Crystallogr. 1983, 39A, 158. (16) Bronger, W.; Bohmer, M.; Schmitz, D. Z Anorg. Allg. Chem. 2000, 626, 6-8.
Mn1 Mn2 Mn3 Mn4 K S1 S2
x
y
z
Ueqa
occupancy
5000 2280(1) 0 2500 1235(2) 1327(1) 1173(2)
2500 2500 3(1) 0 1252(1) 1969(1) 533(1)
2500 2500 7500 -2506(4) 75(2) 5314(3) 4988(3)
7(1) 9(1) 36(1) 17(1) 22(1) 11(1) 24(1)
1.0 1.0 1.0 0.50 1.0 1.0 1.0
a U eq is defined as one-third of the trace of the orthogonalized Uij tensor.
material using conventional serial four-circle diffractometers. Only the cell parameters of K2Mn3S4 could be determined which were the same as those reported by Bronger. Though data were collected with a four-circle diffractometer, the structure could not be solved in a satisfactory manner, partly because of the broad and weak nature of the diffraction peaks and partly because of the Mn atom disorder. During the course of this work the Department of Chemistry was equipped with a Siemens SMART Platform CCD diffractometer which proved to be crucial in eventually determining the structure of K2Mn3S4. This is because weak reflections could be better measured and we were able to observe a considerably larger orthorhombic cell (4× larger than the monoclinic one); see Table 3. A full reciprocal lattice sphere of data was recorded to a resolution of 0.75 Å. Three major swaths of frames were collected with 0.30° steps in ω and an exposure time of 20 s/frame, giving a total measurement time of 0.6 samples, the values of C and θ obtained from the high-temperature data continue to increase; see Figure 14B. The increasing θ values indicate further increases in the strength of the antiferromagnetic coupling, as expected from the increasing concentration of Mn centers in the sample. However, given that it is questionable whether CW behavior is
still in effect in these samples, any attempt to extract a C and θ is of limited value. The shape of the curve for the end member K2Mn3S4 is very similar to that of Cs2Mn3S4.27,28 As the concentration of Mn increases, the susceptibility deviates from CW behavior in ways that depend on the various Mn-S-Mn super-exchange interactions. At high Mn concentrations, one may expect a magnetic phase transition. Some possible transitions are from paramagnetic (P) behavior to a long-range ordered ferromagnetic (F), antiferromagnetic (AF), or ferrimagnetic (FI) state. Another possibility is a transition from P behavior to a short-range spin glass (SG) state. In a SG state, the spins are located randomly on structurally disordered sites so that the exchange interactions are “in conflict.” At the freezing temperature, Tf, the spins freeze into what appear to be random directions but are in fact the result of the system attempting to resolve the conflict. A signature of the P to SG transition is a maximum in the χm vs. T curve measured when the sample is cooled in zero field (ZFC), with somewhat different behavior observed if the sample is cooled in a field (FC). Which of these phases actually occurs depends on the sign of the nearest neighbor and the next nearest neighbor exchange interactions as well as the disorder and subsequent competition between these (28) K2Mn3S4 appears to have complex magnetic behavior which includes a ferromagnetic transition at 44 K. The properties of this material are outside the scope of this paper. Hanko, J.; Kanatzidis, M. G., unpublished results.
2862
Chem. Mater., Vol. 13, No. 9, 2001
Axtell et al.
Figure 16. (A) Temperature-dependent, in-phase magnetization for K2Cd1.2Mn1.8S4 (x ) 0.6) at various driving frequencies. (B) The field dependence of the magnetization for several K2Cd3(1-x)Mn3xS4 solid solutions (Inset: the low field region of the x ) 0.6 sample).
interactions. (There exist so-called “re-entrant spin glass” systems that exhibit two successive phase transitions with decreasing temperature, a P to F transition followed by a SG transition). It is at x ) 0.6 that a transition to an ordered magnetic phase is first observed with differences between the ZFC data and the FC data, see Figure 15. This behavior is somewhat suggestive of spin glass behavior. In Figure 15A, the data were obtained with a measuring field of 100 G. The location of the peak in the ZFC data is dependent on the magnitude of the applied field for this composition; see Figure 15B. With increasing applied field, the peak broadens and becomes weaker. The temperatures of the peaks are 18, 15, and 9 K for the runs at 10, 100, and 400 G, respectively. At 4000 G, the peak is washed out or it occurs below 5 K. This behavior, too, is reminiscent of a spin glass where, with increasing field, one must go to lower temperatures to achieve freezing of the spins in random orientations. The x ) 0.87 phase also shows substantial differences between the ZFC data and the FC data. Two peaks are observed in the ZFC data, at 14 and 39 K; see Figure 15C. These data are consistent with spin-glass-like behavior as well; however, the existence of two peaks is puzzling. Possible reasons for the two peaks may be the presence of two different phases or the presence of crystals with two different concentrations of Mn (e.g., concentration gradients). None of these possibilities
Figure 17. (A) Hysteresis behavior of K2Cd1.2Mn1.8S4 (x ) 0.6). (B) Hysteresis behavior of K2Cd0.39Mn2.61S4 (x ) 0.87).
could be distinguished from the XRD powder patterns or microprobe EDS analyses of these samples. This behavior, however, may also be due to a single phase because it was reproducible from batch to batch. The locations of the two peaks are dependent on the applied field in this sample as well; see Figure 15D. The higher temperature peak has values of 35, 34, and 33 K for H ) 10, 100, and 500 G, respectively. The lower temperature peak occurs at 17, 15, and 10 K for H ) 10, 100, and 500 G, respectively. The temperature-dependent ac magnetization was determined for the x ) 0.6 and x ) 0.87 samples.29 The x ) 0.6 sample showed two peaks, at 9 and 21 K, and no frequency dependence; see Figure 16. The in-phase and out-of-phase data sets contained peaks at the same temperatures, varying only in intensity. For K2Cd0.39(29) In an ac measurement, one observes both the in-phase and outof-phase components of the magnetization. In P, F, AF, and FI systems, where there is no serious time dependence of the magnetization, and there are no interesting features in the ac magnetization. In SG systems, where the magnetization varies logarithmically with time, the peak in the magnetization versus temperature curve occurs at a freezing temperature, Tf, which decreases with increasing measuring time (reciprocal of the measuring frequency) approaching the true phase transition temperature Tsg in the limit of infinite measuring time.
Photoluminescence and Magnetism in K2Cd3(1-x)Mn3xS4
Chem. Mater., Vol. 13, No. 9, 2001 2863
Mn2.61S4 (x ) 0.87), a much weaker response was observed. Two peaks, at 18 and 35 K, were observed with driving frequencies of 1 and 10 Hz. At 100 and 1000 Hz, the data sets were too noisy to be reliable. It is interesting to note that the peaks for the x ) 0.87 material seem to agree with the two peaks in the dc data, as does the high-temperature peak in the x ) 0.6 sample. Hysteresis. To further assess any spin-glass-like behavior, we performed field-dependence measurements in field-cooled and zero-field-cooled samples. In a spin glass, the shape and position of the loop may depend on the field in which the sample was cooled prior to “cycling”. For each composition, the field dependence at 5 K was determined. None of the samples appeared to reach a saturation field; see Figure 17A. The x ) 0.6 sample, however, displayed an inflection point in the data; see the inset in Figure 17. The decreasing values of the magnetization with increasing Mn content suggest that at higher x values larger percentages of the manganese atoms are paired antiferromagnetically, consistent with the increasingly negative θ values. This could be due to the formation of antiferromagnetic dimers which “drop out” of the magnetization. The hysteresis behavior was investigated for the x ) 0.4, 0.6, and 0.87 samples. The 0.4 sample shows no hysteresis at all, while the 0.6 sample shows significant hysteresis; see Figure 17A. For the x ) 0.6 sample, a remnant magnetization of 125 emu/mol Mn was observed at 0-G applied field, and a coercive field of 413 G. It is interesting to note that the hysteresis loop does not attain saturation up to 30 000 G and this could be a further indication of spin-glass behavior in this system. The hysteresis behavior of K2Mn2.61Cd0.39S4 (x ) 0.87) was much weaker than that of the x ) 0.6 sample. The remnant magnetism is 4.3 emu/mol Mn, while the coercive force is 195 G; see Figure 17B. The shape of the hysteresis loop is also poor and the loop never really closes on itself because saturation is not achieved up to the highest available field. The smaller dimensions of the hysteresis loop compared to the x ) 0.6 sample indicate both that a smaller number of spins remain aligned at zero field and that a smaller opposing field is necessary to reverse the spins. These effects could be due to the increase in the manganese content and/or the structural change in this sample. The above magnetic data may indicate the presence of finite length chains (clusters) of manganese atoms, running in a zigzag manner through the layers. The chains are expected to possess antiferromagnetic coupling between manganese atoms.30 The end members of odd numbered chains however may contribute to paramagnetic behavior. With increasing amounts of manganese, the chains lengthen, increasing the amount of antiferromagnetic behavior and decreasing the number of lone paramagnetic manganese atoms. What remains to be discerned is how these chains interact with each other to produce the ordering observed in the temperature dependence and hysteresis data. Chains of metal atoms in the same layer, spaced by a cadmium atom, could interact ferromagnetically. These interac-
tions are expected to be weak because they are next nearest neighbor interactions. Between layers metal atoms are connected by a M-S-K-S-M sequence. The exchange is expected to be weak here, too, because the K-S bonds are long and possess high ionic character. A theoretical calculation of the magnetic susceptibility for the K2Cd3(1-x)Mn3xS4 solid solutions will be complicated both by the highly anisotropic nature of these layered materials and the complicated exchange geometry within each layer. Although we have pointed out that at high x values the magnetic properties are “spinglass-like”, we cannot confidently state that they are in fact spin glasses because they do not possess all the attributes associated with such behavior (e.g., lack of ac frequency dependence, etc.). It is possible that some of these observations may also partially fit the description for canted antiferromagnetism. Additional work is needed to further clarify this issue.
(30) Bronger, W.; Muller, P. J. Alloy Compds. 1997, 246 (1-2), 27-36.
(31) Agekyan, V. F.; Vasil’ev, N. N.; Serov, A. Y. Phys. Solid State 1999, 41, 41-44.
Concluding Remarks A series of congruently melting K2Cd3(1-x)Mn3xS4 solid solutions has been demonstrated. These are magnetic semiconductors which are chemically derived from CdS but with larger energy band gaps than those of the binary sulfide. For x < 0.6, these solid solutions feature (M3S4)n2n- layers assembled only of truncated (M3S4)2cubes. A phase change and a structural reorganization of the layers occurs between 0.6 < x < 0.9. The new layers are similar to those found in Cs2Zn3S4. At x ) 1 the structure changes again, though it remains layered, to the K2Mn3S4 type. The compounds display red emission, even at room temperature, which suggests possible applications as red phosphors. The emission intensity and type depends strongly on x. Small amounts of Mn cause a small bathochromic shift in the emission, but increasing amounts quench the photoluminescence. This is reminiscent of a previous study in which the dependence of luminescence intensity on manganese-ion concentration in the dilute magnetic semiconductors Cd1-xMnxTe with 0.4 < x < 0.731 was shown to saturate because of effective nonlinear quenching already at low excitation levels. The unique, layered structure of the parent compound K2Cd3S4 provides an interesting host lattice for the insertion of magnetic centers such as manganese. The magnetic behavior of these solid solutions is unusual and does not fit well any of the standard magnetism models. Although antiferromagnetic coupling is observed in most cases, with very high Mn content and at low temperatures, transitions to magnetically complex phases are observed. Other divalent magnetic centers such as Co and Fe should also give analogous solid solutions. Acknowledgment. This research was made possible with funding from NSF Grant DMR-9817287. This work made use of a scanning electron microscope located in the Center for Advanced Microscopy at Michigan State University. Supporting Information Available: Crystal data and anisotropic thermal parameters for K2Mn3S4 and K2Cd0.35Mn2.65S4 (PDF). This material is available free of charge via the Internet at http://pubs.acs.org. CM0100422
